{"id":"W1975396736","doi":"10.1109/mdt.2010.73","title":"Time to retire our benchmarks","year":2010,"lang":"en","type":"article","venue":"IEEE Design & Test of Computers","topic":"Distributed and Parallel Computing Systems","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Argument (complex analysis); Computer science; Quarter (Canadian coin); Sentence; Artificial intelligence; History","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0007110311,0.0002560539,0.0003778261,0.0001724422,0.0001039995,0.0001732168,0.002048974,0.0001246244,0.00001481826],"category_scores_gemma":[0.00009569726,0.0002520112,0.0001237611,0.000610586,0.00004065372,0.0002186451,0.0001882175,0.0002714002,0.0005150809],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002220563,"about_ca_system_score_gemma":0.0001337895,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001386025,"about_ca_topic_score_gemma":6.482376e-7,"domain_scores_codex":[0.9980339,0.0001158431,0.0004544603,0.0005267209,0.0004216715,0.0004473756],"domain_scores_gemma":[0.9978797,0.0004741497,0.0002023118,0.000969636,0.0001993663,0.000274795],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003082606,0.0004880443,0.0004085294,0.00008387167,0.00009103363,0.000140023,0.001388587,0.08327226,0.1058643,0.003097945,0.7683045,0.03683006],"study_design_scores_gemma":[0.0008128578,0.001028923,0.001826044,0.0002775186,0.00001734684,0.000233802,0.00001525651,0.9576957,0.0112055,0.0007926132,0.0251201,0.0009742925],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01460704,0.00001455185,0.9786541,0.001023083,0.002450635,0.0003644134,0.000009369659,0.0002817048,0.002595068],"genre_scores_gemma":[0.8616783,4.315687e-7,0.1374587,0.0003125849,0.0002886755,0.000008297876,0.000006456187,0.00001381822,0.0002327882],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8744235,"threshold_uncertainty_score":0.9999932,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01433978754296901,"score_gpt":0.2362975613649401,"score_spread":0.2219577738219711,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}